TheSwamp

CAD Forums => Vertically Challenged => Land Lubber / Geographically Positioned => Topic started by: Mark on August 02, 2007, 03:16:24 PM

Title: ( C3D ) Accumulation errors
Post by: Mark on August 02, 2007, 03:16:24 PM
How do you guys deal with accumulation errors in C3D?

One of my drawings just went down in flames .... over 0.01'. The stations didn't match the inverse distance on my alignment. I'm guessing it's round-off.
Title: Re: ( C3D ) Accumulation errors
Post by: Dinosaur on August 02, 2007, 03:19:54 PM
I haven't tried this but can you set up a label style that truncates the rounding rather than round normal and give that style to the questionable course?  It would have to be done in the label component editor per yesterday's problem.
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 02, 2007, 03:33:24 PM
I'll try tomorrow Dino. I had to explode the stations and renumber them to get it out the door today.
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 02, 2007, 03:47:46 PM
The problem with this alignment was if you added up the total distance using the stations it wasn't the same as the total distance shown by the inverse labels. So C3D is taking the total distance then rounding the numbers. /guess
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 02, 2007, 04:28:10 PM
Plan distances or TRUE distances??
Title: Re: ( C3D ) Accumulation errors
Post by: Dinosaur on August 02, 2007, 05:54:00 PM
Plan distances or TRUE distances??
They would be the real distances - it is just an accumulation of the program rounding one direction than the other.  Civil 3D , like its grandpa Land Desktop, is quit anal about the intermediate course distances being rounded with  exactly the same precision as the overall distance and the nature of an overall design dictates that it is more important that ends of a course fall exactly at an intersection or it runs exactly parallel with or perpendicular to something rather than measure an exact distance or bearing.  This means it is common for a course that says it is 123.45 feet to be actually be 123.4549 feet or 123.4451 feet.  All it takes is for two runs at either extreme to have the sum of the courses be more than the precision level allows and the overall length will show as being 0.01 foot different.  That less than 1/8 inch drives reviewers bat-squack and once found, the design is flagged as inaccurate due to rounding and kicked back for non compliance.  If this is in a critical area that has been designed at minimums, the correction may force slopes for streets or pipes outside of allowable range or cause a lot to not meet area requirements or not close with the required precision.
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 02, 2007, 06:22:39 PM
That less than 1/8 inch drives reviewers bat-squack and once found, the design is flagged as inaccurate due to rounding and kicked back for non compliance.  If this is in a critical area that has been designed at minimums, the correction may force slopes for streets or pipes outside of allowable range or cause a lot to not meet area requirements or not close with the required precision.
Somebody needs smacked with a common sense stick.  How'd they do it thirty years ago??
Title: Re: ( C3D ) Accumulation errors
Post by: Dinosaur on August 02, 2007, 06:31:55 PM
Somebody needs smacked with a common sense stick.  How'd they do it thirty years ago??
with their K&E jewel points on the vellum by hand.  They could make the numbers work.  A Civil 3D label is part of the object itself and as yet, can not have its value over ridden.  You can explode the label and the resulting block and lose all associativity or you can find a way to trick the program to display the numbers you want - like the truncating option I told Mark about which is best if it will work.  He can also shorten or lengthen his alignment just enough to make the overall number match the sums - a problem if he shortens it and he is going to an intersection because he just lost his intersection.
Title: Re: ( C3D ) Accumulation errors
Post by: Jeff_M on August 02, 2007, 07:58:03 PM
It took some doing, but I think I finally have convinced the main architect we work with to quit giving us houses that are XX'- 4" (or 8") wide. They have agreed to go with 0", 3" 6", or 9" because 10 lots 45.33' wide adds up to 453.30' but the overall dimension will always come up 453.33'. Yes, it can be fudged, but everyone's life is easier if you eliminate the potential problem.
Title: Re: ( C3D ) Accumulation errors
Post by: Jeff_M on August 02, 2007, 08:06:56 PM
What I've done is create Label styles that use Expressions. The Expression formula is merely {Segment Length} - 0.01.  The label itself I set to color RED so it stands out from the regular labels and I can quickly tell which labels have been altered. This way, when linework changes I can reset the label to a normal label and check for the accumulation error again.
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 08:55:23 AM
I haven't tried this but can you set up a label style that truncates the rounding rather than round normal and give that style to the questionable course?  It would have to be done in the label component editor per yesterday's problem.

Truncating the Geometry Point label almost worked.
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 08:58:10 AM
That less than 1/8 inch drives reviewers bat-squack and once found, the design is flagged as inaccurate due to rounding and kicked back for non compliance.  If this is in a critical area that has been designed at minimums, the correction may force slopes for streets or pipes outside of allowable range or cause a lot to not meet area requirements or not close with the required precision.
Somebody needs smacked with a common sense stick.  How'd they do it thirty years ago??

Like I did it yesterday .... by hand! Of course I might as well be using vanilla ACAD if we have to keep labeling every thing by hand.
Title: Re: ( C3D ) Accumulation errors
Post by: Dinosaur on August 03, 2007, 08:59:56 AM
Jeff probably had the best suggestion unless tweaking the endpoint of your alignment won't cause you problems.
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 03, 2007, 09:03:11 AM
That less than 1/8 inch drives reviewers bat-squack and once found, the design is flagged as inaccurate due to rounding and kicked back for non compliance.  If this is in a critical area that has been designed at minimums, the correction may force slopes for streets or pipes outside of allowable range or cause a lot to not meet area requirements or not close with the required precision.
Somebody needs smacked with a common sense stick.  How'd they do it thirty years ago??

Like I did it yesterday .... by hand! Of course I might as well be using vanilla ACAD if we have to keep labeling every thing by hand.

I guess my point was the numbers were the very same (round-off and all) but no one had a canary over and 1/8" on an alignment.  The surveyors won't get it that close three times in a row over any real distance.

Can you change to three decimal places instead of two?
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 09:14:37 AM
Jeff probably had the best suggestion unless tweaking the endpoint of your alignment won't cause you problems.
Only you can't edit individual geometry points on an alignment, that I know of anyway. I could tweak the endpoint but then the geometry in the dwg would be wrong, which will likely mean more to the person that receives it.
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 09:21:11 AM
That less than 1/8 inch drives reviewers bat-squack and once found, the design is flagged as inaccurate due to rounding and kicked back for non compliance.  If this is in a critical area that has been designed at minimums, the correction may force slopes for streets or pipes outside of allowable range or cause a lot to not meet area requirements or not close with the required precision.
Somebody needs smacked with a common sense stick.  How'd they do it thirty years ago??

Like I did it yesterday .... by hand! Of course I might as well be using vanilla ACAD if we have to keep labeling every thing by hand.

I guess my point was the numbers were the very same (round-off and all) but no one had a canary over and 1/8" on an alignment.  The surveyors won't get it that close three times in a row over any real distance.

Can you change to three decimal places instead of two?
No they won't and we know that, but the person checking the drawing will add up all the numbers it they won't match, red flags will ensue.

No, two places it what I have to use, boss says so.
Title: Re: ( C3D ) Accumulation errors
Post by: Dinosaur on August 03, 2007, 09:43:04 AM
Only you can't edit individual geometry points on an alignment, that I know of anyway. I could tweak the endpoint but then the geometry in the dwg would be wrong, which will likely mean more to the person that receives it.
How much overall length error is there?  With only .01 accumulated error, fudging by .005 in the right direction should be more than enough to make the sums total to the overall.  The only downside is if you have to reduce the length and you are ending at an intersection you don't have the intersection any more.  You shouldn't need to change any intermediate points at all.

I agree Randy, it is total insanity . . . we are talking about less than half the diameter of the bars they are setting for this stuff IF the hit it exactly.

A few years back we had a complicated plat surrounded by established plats, a section line and right-of-way.  None of the boundary could be fudged but it wouldn't close within acceptable error.  We finally had to take the boundary down to the thousandths and call out deed and measured distances for every course to prove closure.
Title: Re: ( C3D ) Accumulation errors
Post by: mjfarrell on August 03, 2007, 09:45:02 AM
I have previously reported this issue through channels.  Only to have the import of be dismissed;
Those error are 'only' happening at say 5 decimal places, they aren't using it to that accuracy are they?

Well duh!


It is an Engineering application and IF we can't trust the numbers coming out of it, and or they are consistently wrong........

Perhaps testing of the product should involve the accuracy of the results, not does it NOT crash when said function is used.
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 09:51:17 AM
What I've done is create Label styles that use Expressions. The Expression formula is merely {Segment Length} - 0.01.  The label itself I set to color RED so it stands out from the regular labels and I can quickly tell which labels have been altered. This way, when linework changes I can reset the label to a normal label and check for the accumulation error again.

Actually this is the only way to make it work. Thanks Jeff.

I was trying to make the stations match the inverse distance but that doesn't look like it's going to work so .... fudge it!! :)
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 10:32:09 AM
What I've done is create Label styles that use Expressions. The Expression formula is merely {Segment Length} - 0.01.  The label itself I set to color RED so it stands out from the regular labels and I can quickly tell which labels have been altered. This way, when linework changes I can reset the label to a normal label and check for the accumulation error again.

Actually this is the only way to make it work. Thanks Jeff.

I was trying to make the stations match the inverse distance but that doesn't look like it's going to work so .... fudge it!! :)

OK .... well that doesn't work either. I still can't get the stations to match the inverse distance or vice-versa. Tried changing the rounding of both labels but I can't make it work.
Title: Re: ( C3D ) Accumulation errors
Post by: mjfarrell on August 03, 2007, 10:39:03 AM
I think one could cheat, by doing what DinØ did.
Draw the pipe alignment, and then reference that alignment for all station, length information. Thus the pipes are like supermodels, pretty to look at, but none too smart.

Is your Software smarter than a 5Th Grader?
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 03, 2007, 11:33:21 AM
I think one could cheat, by doing what DinØ did.
Draw the pipe alignment, and then reference that alignment for all station, length information. Thus the pipes are like supermodels, pretty to look at, but none too smart.

Is your Software smarter than a 5Th Grader?
The software is doing EXACTLY what you're telling it to do.  That's the way math works, if you "round-off" to 2 decimal places the result is LESS accurate.  Open up any spreadsheet tool you'd like and add 1.49 and 1.49, then set the round-off to zero decimal places, it'll tell you that 1+1=3.  There is over 1/32" difference between 0.33' and 4", which do you want? What else would you have it do??

The problem is a combination of some checker unable to understand rounding, (posessing not enough common sense to let the 1/8" fly within acceptable tolerances), and a boss locked into two decimal places for probably no more reason than "that's how we've always done it".
Title: Re: ( C3D ) Accumulation errors
Post by: mjfarrell on August 03, 2007, 11:44:02 AM
uh huh, only the software is 'forcing' us to round off, because, unless ones pipe is totally straight it will report the incorrect distance even if you do not round off or truncate the pipe 2d distance.  The software fails to report true 2d distance if the pipe isn't laser level in the ground and we all know that won't get approved for sure. :evil:
Title: Re: ( C3D ) Accumulation errors
Post by: Jeff_M on August 03, 2007, 12:47:45 PM
The problem is a combination of some checker unable to understand rounding, (posessing not enough common sense to let the 1/8" fly within acceptable tolerances), and a boss locked into two decimal places for probably no more reason than "that's how we've always done it".
Around here it's not just the bosses. I went round and round with a plan checker where a group of lot lines's frontage lengths did not add up to the total Right of Way line length. I'd do what they asked, namely fudge a number, but now the labeled distance didn't match the lot closure calcs. Finally, I submitted the entire plat to 3 decimal places.

My office is about 8 miles from City Hall......we could hear the explosion of their collective heads when they received THAT plan for checking. My boss got a phone call within minutes to get down there and pick up the plan and get it revised back to standards and to NEVER try that "stunt" again.

It really is a no-win situation when working with Government agencies.
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 03, 2007, 01:05:26 PM
It really is a no-win situation when working with Government agencies.
We're back to smacking somebody with a "common sense" stick .... very very hard.  We spend lots of money buying accurate tools.  Then using those tools we work very hard producing an accurate representation.  Then some twit (with stroke) wants to break that accuracy because the drawing doesn't look the way it did thirty-five years ago.  Ya' know, I've heard that "common sense" sticks now come in aluminum, a little lighter with a pleasent sort of "ring" to it after impact.
Title: Re: ( C3D ) Accumulation errors
Post by: Dinosaur on August 03, 2007, 01:33:15 PM
It appears that the design submittals are no longer being checked for engineering accuracy, but only for visually confirming that the plans conform to the current criteria and specifications and the actual design is assumed to be correct because a computer generated it.  A string of numbers not totaling to the overall is a red flag to that type of review.  In fact, it is very rare to actually have an engineer review our plans at least on a first submittal as budget cuts weed out professional staff leaving a clerk with a "go by" set of approved plans to compare with.  Only in the final review does a major design issue get discovered when it has a much greater impact on budgets and schedules.
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 01:50:57 PM
The software is doing EXACTLY what you're telling it to do.  That's the way math works, if you "round-off" to 2 decimal places the result is LESS accurate.  Open up any spreadsheet tool you'd like and add 1.49 and 1.49, then set the round-off to zero decimal places, it'll tell you that 1+1=3.

Not EXACTLY.
http://www.theswamp.org/screens/mark/screen_shots/stations.png
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 02:01:21 PM
We spend lots of money buying accurate tools.
Don't get me started on that one again .......... :)

Quote
Then using those tools we work very hard producing an accurate representation.
WRONG! ( http://www.theswamp.org/screens/mark/screen_shots/stations.png ) isn't an accurate representation.

Quote
Then some twit (with stroke) wants to break that accuracy because the drawing doesn't look the way it did thirty-five years ago.
In my case it IS accuracy. If the numbers don't add up then there is SOMETHING wrong with the drawing. "Looks" didn't even come into play but I know what you're talking about. "Been there done that" as they say.
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 03, 2007, 02:15:16 PM
A string of numbers not totaling to the overall is a red flag to that type of review. 
And someone with a little sense would look at the numbers and understand that an 1/8" over a couple hundred feet was a rounding adjustment, not an error that required flagging

Only in the final review does a major design issue get discovered when it has a much greater impact on budgets and schedules.
Burning hours worrying about how it "looks", but nothing about whether it "works".  I've even heard of some "common sense" sticks that come in calibers and full-auto.
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 03, 2007, 02:25:37 PM
Quote
Then using those tools we work very hard producing an accurate representation.
WRONG! ( http://www.theswamp.org/screens/mark/screen_shots/stations.png ) isn't an accurate representation.
.....
In my case it IS accuracy. If the numbers don't add up then there is SOMETHING wrong with the drawing.
There is rounding going on there (rounding is not inaccurate, its just rounding). Run all the numbers out to eight decimal places.  Even with the posted numbers you are quite accurate to within a hundreth of a foot, less than an eighth of an inch over six hundred feet.
Title: Re: ( C3D ) Accumulation errors
Post by: Dinosaur on August 03, 2007, 02:27:47 PM
But Randy,  this is municipal review staff we are tying to get past to have a project approved to join the public infrastructure.  The dirty little secret is that sometimes the city or county does not really want something built there yet because some facet of city service won't be adequate to absorb the new development.  Regardless, the project can not go forward until each reviewer involved is convinced all of the information is complete and correct by whatever criteria they are using.  The only control we can exert is send the client to someone's boss to get the bearings lubricated.  If the client has enough clout it even works on occasion.
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 03, 2007, 02:35:49 PM
But Randy,  this is municipal review staff we are tying to get past to have a project approved to join the public infrastructure.  The dirty little secret is that sometimes the city or county does not really want something built there yet because some facet of city service won't be adequate to absorb the new development.  Regardless, the project can not go forward until each reviewer involved is convinced all of the information is complete and correct by whatever criteria they are using.  The only control we can exert is send the client to someone's boss to get the bearings lubricated.  If the client has enough clout it even works on occasion.
Oh I quite understand that.  Such is not a problem with the tool however.  We're back to the stick. 

Rounding discrepancies (not errors) are the primary reason I’ll never dual dimension Imperial/Metric on a drawing.  Everything on the drawing is quite accurate, but rounding pushes some dimension a millimeter or a sixteenth one way or the other, and some bozo checker somewhere “knee-jerks” into slobbering hysteria.  “Bozo, meet stick.”  My, that is a pleasant “ring” isn’t it?
Title: Re: ( C3D ) Accumulation errors
Post by: Mark on August 03, 2007, 03:02:33 PM
Quote
Then using those tools we work very hard producing an accurate representation.
WRONG! ( http://www.theswamp.org/screens/mark/screen_shots/stations.png ) isn't an accurate representation.
.....
In my case it IS accuracy. If the numbers don't add up then there is SOMETHING wrong with the drawing.
There is rounding going on there (rounding is not inaccurate, its just rounding). Run all the numbers out to eight decimal places.
If I show a inverse distance between two stations and it doesn't match the station distance then it is not accurate. Why does C3D ( the "tool" ) do that? It's labeling the stations AND the inverse distance of that segment!
Title: Re: ( C3D ) Accumulation errors
Post by: CADaver on August 03, 2007, 03:32:10 PM
Quote
Then using those tools we work very hard producing an accurate representation.
WRONG! ( http://www.theswamp.org/screens/mark/screen_shots/stations.png ) isn't an accurate representation.
.....
In my case it IS accuracy. If the numbers don't add up then there is SOMETHING wrong with the drawing.
There is rounding going on there (rounding is not inaccurate, its just rounding). Run all the numbers out to eight decimal places.
If I show a inverse distance between two stations and it doesn't match the station distance then it is not accurate.
It is quite accurate within the parameters you've set, 2 decimal places.
Title: Re: ( C3D ) Accumulation errors
Post by: Jeff_M on August 03, 2007, 03:45:53 PM
It's labeling the stations AND the inverse distance of that segment!
No, it's labeling the stations based on the position of the labeled points along the alignment in relation to the beginning station of the alignment. The labeled segment is just that, a segment whose length is not governed by the total length, but is a part of the overall length.

This is something that has boggled many minds when working in the City I mainly work in. The guys checking the Road plans seem to have a grasp on reality and have never questioned the Station Difference vs segment length. Possibly because we supply them with centerline calcs output to 3 decimal places, including the stations, and they can see for themselves that it is a rounding issue. I just wish they'd talk to the guys doing the map checking, because they don't have a clue.....
Title: Re: ( C3D ) Accumulation errors
Post by: jpostlewait on August 03, 2007, 10:10:56 PM
I'll try tomorrow Dino. I had to explode the stations and renumber them to get it out the door today.

So you are done and you move on.
If you think it's nuts to sit around and argue about hundredths of a foot with an idiot, take him out to the job site and tell him you set one hub a hundredth incorrectly and ask him to point it out.
For a guy how grew up in the field the inane math requirements placed upon plan documents either civil or building is crazy.
One of our DOT's requires that all Math be precise at two decimal places. So you just construct everything to exactly 2 decimal place distances.
All slopes are evenly divisible by 4.
I know why this made sense 40 years ago.
Some rates of change are slower than others.